Disseminated cytomegalovirus (CMV) infection is a major cause of morbidity and mortality in AIDS patients. Most HIV-infected persons have previously been infected with CMV; 25% may experience life or sight threatening consequences. Our long therm goal is to determine how HIV- infected persons at risk for CMV-related morbidity can be identified. We hypothesize that persons at greatest risk of CMV disease are more likely to carry replicating CMV in peripheral blood leukocytes (PBL), carry a greater viral load, or carry CMV in a particular subpopulation of PBL. We will examine differences among HIV infected persons in presence and quantity of replicating or nonreplicating CMV in PBL, or in specific subpopulations of PBL. We will study persons who are asymptomatic, asymptomatic but with CD4 counts <400, with AIDS (with and without opportunistic infections), and with CMV-related diseases. At enrollment, two additional visits, time of decrease in immune function or progression to AIDS, and occurrence of CMV disease we will detect CMV in mononuclear and polymorphonuclear fractions of PBL. We will detect presence of CMV genome in PBL and determine presence and quantity of replicating virus by viral culture, detection of late antigen using monoclonal antibody staining, and detection of late RNA using in situ hybridization. We will detect and quantitate nonreplicating CMV by detection of immediate early (IE) antigen using antibody staining, and IE RNA using in situ hybridization. We will estimate how results differ among individuals and in relation to stage of HIV infection and degree of immune dysfunction. We will evaluate the ability of flow cytometry to provide results comparable to those found with other methods of CMV detection. The observations we make will enable us to define a cohort of persons who differ in manifestations of CMV in PBL for a future prospective study to correlate differences with risk for CMV-related morbidity.